Macrophages protect against sensory axon loss in peripheral neuropathy

Abstract

Peripheral neuropathy is a common complication of type 2 diabetes, which is strongly associated with obesity¹, causing sensory loss and, in some patients, neuropathic pain^2,3. Although the onset and progression of diabetic peripheral neuropathy is linked with dyslipidaemia and hyperglycaemia⁴, the contribution of inflammation to peripheral neuropathy pathogenesis has not been investigated. Here we used a high-fat, high-fructose diet (HFHFD), which induces obesity and prediabetic metabolic changes, to study the onset of peripheral neuropathy. Mice fed the HFHFD developed persistent heat hypoalgesia after 3 months, but a reduction in epidermal skin nerve fibre density manifested only at 6 months. Using single-cell sequencing, we found that CCR2⁺ macrophages infiltrate the sciatic nerves of HFHFD-fed mice well before axonal degeneration is detectable. These infiltrating macrophages share gene expression similarities with nerve-crush-induced macrophages⁵ and express neurodegeneration-associated microglial marker genes⁶, although there is no axon loss or demyelination. Inhibiting the macrophage recruitment by genetically or pharmacologically blocking CCR2 signalling resulted in more severe heat hypoalgesia and accelerated skin denervation, as did deletion of Lgals3, a gene expressed in recruited macrophages. Recruitment of macrophages into the peripheral nerves of obese prediabetic mice is, therefore, neuroprotective, delaying terminal sensory axon degeneration by means of galectin 3. Potentiating and sustaining early neuroprotective immune responses in patients could slow or prevent peripheral neuropathy.

Fig. 1. A HFHFD leads to peripheral neuropathy.

a, Illustration of the HFHFD administered to induce prediabetes versus the CD. b, Mechanical withdrawal thresholds of male mice fed a HFHFD or CD over time using an up–down von Frey method (n = 8 mice per group; data are presented as mean values ± s.e.m., P values from two-way analysis of variance (ANOVA) with Sidak multiple comparisons). c, Heat withdrawal latency in Hargreaves assays of male mice fed HFHFD or CD (n = 8 mice per group; data are presented as mean values ± s.e.m., P values from two-way ANOVA with Sidak multiple comparisons). d, Representative heat maps showing the location of a mouse during a 1-h thermal gradient ring recording. e, Time spent on a thermal gradient ring by male mice fed a HFHFD or CD for 12 weeks (n = 8 mice per group; data are presented as mean values ± s.e.m. (shaded area); P = 0.050194 (10 °C), 0.052991 (14 °C), 0.011189 (17 °C), 0.026107 (21 °C), 0.006838 (24 °C), 0.456255 (28 °C), 0.004662 (31 °C), 0.234499 (35 °C), 0.053147 (39 °C), 0.340637 (45 °C), 0.038695 (51 °C) and 0.322922 (56 °C), from multiple unpaired Mann–Whitney tests, *P < 0.05, **P < 0.005). f, Representative PGP9.5 staining of hind paw skin of male mice fed a HFHFD or CD for 24 weeks showing axon terminals in the epidermis marked by PGP9.5 staining and cell layers by 4′,6-diamidino-2-phenylindole (DAPI) staining. Scale bars, 50 μm. g,h, Quantification of IENFD (left) and CGRP⁺ fibres (right) at 12 weeks of diet (n = 7 (CD) and 8 (HFHFD) for PGP9.5, and n = 9 per group for CGRP; data are presented as mean values ± s.e.m., P values from two-tailed unpaired Student’s t-test; g) and 24 weeks of diet (n = 7 mice per group; data are presented as mean values ± s.e.m., P values from two-tailed unpaired Student’s t-test; h). The graphics in a were created with BioRender.com. Source Data

Fig. 2. CCR2⁺ macrophages infiltrate sciatic nerves of mice fed HFHFD and resMacs become pro-inflammatory.

a, Uniform manifold approximation and projection (UMAP) plot showing clusters identified in sciatic nerves from mice fed a CD or HFHFD for 12 weeks (2 samples per group). DCs, dendritic cells; ILCs, innate lymphoid cells; snMacs, sciatic nerve macrophages. b, Bar plot of cell proportions showing increase in CCR2⁺ macrophages in sciatic nerves of HFHFD- versus CD-fed mice. c, Representative cross-section image of sciatic nerves from mice fed a CD or HFHFD for 12 weeks. Col IV, collagen type IV. Scale bars, 100 μm. d, Quantification of F4/80⁺ macrophages in the endoneurium from nerves of mice fed a CD or HFHFD for 12 weeks (n = 5 mice per group; data are presented as mean values ± s.e.m., P value from two-tailed unpaired Student’s t-test). e, Flow cytometry quantification from sciatic nerves of mice fed a CD or HFHFD for different durations, assessing recMacs (LY6C^hiCD64⁺) in the CD11B⁺CD45⁺ live-cell gate (n = 6 mice per group at 8 weeks, 7 mice per group at 10 weeks, and 7 mice in CD and 8 mice in HFHFD at 12 weeks; data are presented as mean values ± s.e.m., P values from two-way ANOVA with Sidak multiple comparisons). f, Volcano plot of gene markers differentially expressed in sciatic nerve resMac (CSF1R⁺CCR2⁻) clusters in HFHFD-fed compared to CD-fed mice, highlighting chemokines and cytokines expressed at higher levels; red indicates genes with labels (encoded protein designations) included in the plot. g, Gene Ontology enrichment analysis of significantly (P > 0.05) upregulated genes from resMac clusters; bold indicates chemokine signaling. h, Representative images from sciatic nerves of CCL2–RFP mice fed a CD or HFHFD for 12 weeks. Scale bars, 100 μm. i, RT–qPCR quantification of Ccl2 expression in nerves from mice fed a HFHFD or CD for 8 or 12 weeks (n = 3 mice per group at 8 weeks, and 3 mice in CD group and 4 mice in HFHFD group at 12 weeks; data are presented as mean values ± s.e.m., P values from two-way ANOVA with Sidak multiple comparisons). Arrowheads indicate F4/80⁺ macrophages that co-express CCL2–RFP; arrows indicate F4/80⁺ macrophages that do not express CCL2–RFP. Source Data

Fig. 3. Macrophages recruited into nerves of mice fed HFHFD for 12 weeks are similar to those recruited days after a traumatic sciatic nerve injury.

a, Integration of macrophages from our HFHFD dataset and a published nerve crush dataset. b, Cell proportions under the different conditions in the integrated datasets. c, Principal component (PC) analysis plot of cell proportions. d, UMAP plots of the merged dataset split by condition. e, Volcano plot of gene markers differentially expressed between CCR2⁺ macrophages (recMacs) and CCR2⁻ macrophages (resMacs). f,g, Violin plots of expression levels of Lgals3 (f) and Trem2 (g) in the merged macrophage datasets. h, Representative flow cytometry plots (left) of recMacs (RMs) (CD64⁺LY6C⁺) from the crushed (ipsilateral) and control (contralateral) sciatic nerves pre-gated on live CD45⁺CD11B⁺LY6G⁻ myeloid cells with quantification (right; n = 4 mice per group; data are presented as mean values ± s.e.m., P value from two-tailed unpaired Student’s t-test). Numbers in outlined areas (left) indicate percent cells in each. i,j, RT–qPCR quantification of Lgals3 (i) and Trem2 (j) expression from nerves of mice fed a CD or HFHFD for 12 weeks or CD mice 1 day post sciatic nerve crush (n = 3 mice (CD), 5 mice (HFHFD) and 4 mice (crush group); data are presented as mean values ± s.e.m., P values from Brown–Forsythe and Welch one-way ANOVA, Dunnett T3 multiple comparison test). The graphics in a were created with BioRender.com. Source Data

Fig. 4. CCR2⁺ macrophages are neuroprotective by means of Gal-3 and blocking their recruitment accelerates peripheral neuropathy.

a, Illustration of experiment. b, Quantification of flow cytometry data of recMacs (LY6C⁺Cd64⁺CD11B⁺) in sciatic nerves of male Ccr2-KO and WT mice fed the HFHFD for 12 weeks (n = 5 mice per group; data are presented as mean values ± s.e.m., P value from two-tailed unpaired Student’s t-test). c, Heat withdrawal latencies in Hargreaves assays of male Ccr2-KO and WT mice fed HFHFD (n = 5 mice per group; data are presented as mean values ± s.e.m., P values from two-way ANOVA with Sidak multiple comparisons). d, Representative PGP9.5 staining of hind paw skin (n = 5 mice per group; scale bars, 50 μm). e, Quantification of IENFD of male Ccr2-KO and WT mice fed a HFHFD for 12 or 24 weeks (n = 5 mice per group; data are presented as mean values ± s.e.m., P values from two-way ANOVA with Sidak multiple comparisons). f, Timeline of CCR2 and CCR5 inhibitor, CVC, injections starting 8 weeks on HFHFD. g, Heat withdrawal latencies in Hargreaves assays of CVC- and vehicle-injected male mice fed HFHFD (n = 9 mice (vehicle) and 8 mice (CVC) group; data are presented as mean values ± s.e.m., P values from two-way ANOVA with Sidak multiple comparisons). h, Quantification of IENFD of CVC- and vehicle-injected male mice fed HFHFD at 12 weeks (n = 6 mice (vehicle) and 4 mice (CVC); data are presented as mean values ± s.e.m., P value from two-tailed unpaired Student’s t-test). i, Illustration of experiment. j, Heat withdrawal latencies in Hargreaves assays of WT and Lgals3-KO mice fed HFHFD for 12 weeks (n = 10 mice per group; data are presented as mean values ± s.e.m., P value from two-tailed unpaired Student’s t-test). k, Quantification of IENFD of WT and Lgals3-KO mice fed HFHFD for 12 weeks (n = 9 mice (CD) and 10 mice (HFHFD); data are presented as mean values ± s.e.m., P value from two-tailed unpaired Student’s t-test). The graphics in a,i were created with BioRender.com. Source Data

Extended Data Fig. 1. High Fat High Fructose feeding leads to obesity and pre-diabetes.

a. Body weights of male C57Bl6 mice fed a HFHFD or CD over six months (n = 8 mice per group, data are presented as mean values +/- SEM, p-value from 2way ANOVA). b. Glucose tolerance test after fasting in male mice fed a HFHFD or CD for 12 weeks. (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from 2way ANOVA). c. Area under the curve from a glucose tolerance test at 12 weeks (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). d. Fasting plasma insulin measurements from male mice fed a HFHFD or CD for 12 weeks (n = 5 mice in CD and 4 mice in HFHFD, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). e. Percent glycated HbA1c measurements from whole blood of male mice fed a HFHFD or CD for 12 weeks (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). f. Glucose tolerance test after fasting in male mice fed a HFHFD or CD for 24 weeks (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from 2way ANOVA). g. Area under the curve from glucose tolerance test at 24 weeks (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). h. Fasting insulin measurement from plasma of male mice fed HFHFD or CD for 24 weeks (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). i. Percent glycated HbA1c measurement from whole blood of male mice fed HFHFD or CD for 24 weeks (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). Source Data

Extended Data Fig. 2. Female mice on a HFHFD are equally susceptible to weight gain, behavioral changes, and immune cell infiltration in nerves.

a. Body weight measurements over time on a HFHFD vs CD from female C57Bl6 mice. (n = 8 mice per group, data are presented as mean values +/- SEM, p-value < 0.0001 from 2way ANOVA) b. Von Frey up down method was used to assess mechanical sensitivity in female mice fed a HFHFD vs CD over 12 weeks. (n = 8 mice per group, data are presented as mean values +/- SEM, p-values from 2way ANOVA with Sidak multiple comparisons) c. Hargreaves assay was used to assess heat sensitivity in female mice fed HFHFD vs CD over 12 weeks. (n = 8 mice per group, data are presented as mean values +/- SEM, p-values from 2way ANOVA with Sidak multiple comparisons) d. Proportions of Ly6c+ CD64+ recruited macrophages from Live Cd45+ Cd11b+ cells in sciatic nerves of female mice fed a HFHFD vs CD for 12 weeks. (n = 8 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). Source Data

Extended Data Fig. 3. Sarm1 knockdown prevents the loss of heat sensitivity and skin axonal degeneration in HFHFD-fed mice but leads to persistent mechanical allodynia.

a. Illustration of the experiment. b. Body weight of WT and Sarm1-KO mice over time on a CD or HFHFD. (n = 5 mice per group, data are presented as mean values +/- SEM). c. Thermal sensitivity in WT vs Sarm1-KO male mice fed a HFHFD or CD over 24 weeks. (n = 5 mice per group, data are presented as mean values +/- SEM, p-values from 2way ANOVA with Sidak multiple comparisons). d. Mechanical sensitivity in WT vs Sarm1-KO male mice fed a HFHFD or CD over 24 weeks. (n = 5 mice per group, data are presented as mean values +/- SEM, p-values from 2way ANOVA with Sidak multiple comparisons). e. Quantification of IENFD from PGP9.5 staining in hind paw skin from WT and Sarm1-KO male mice fed a HFHFD or CD for 24 weeks. (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). f. Summary of findings from Sarm1-KO mouse strain compared to WT mice on a HFHFD. The graphics in a were created with BioRender.com. Source Data

Extended Data Fig. 4. Sciatic nerve immune cell single cell sequencing.

a. Illustration of experimental workflow. b. Number of cells analyzed in each sample post-filtering. c. Dot plot of gene markers used to identify cell types in each cluster. d. UMAP plot showing similar cell composition in replicates of each condition. The graphics in a were created with BioRender.com.

Extended Data Fig. 5. Gating strategy for a flow cytometry assessment of recruited macrophages into the sciatic nerve.

a. FSC-A and SSC-A were used to gate cells and exclude aggregates and small debris. b. FSC-A and FSC-W were used to gate single cells. c. DAPI and CD45-APC-ef780 were used to capture live CD45+ immune cells. d. Cd11b-FITC and Ly6g-PE were used to gate myeloid cells and exclude neutrophils. e. Cd64-PE-594 and Ly6c-BV711 were used to identify Ly6C+ Cd64+ recruited macrophages. (n = 5 mice per group).

Extended Data Fig. 6. Relationship of recruited macrophages to vasculature and identification of source of CCL2.

a. Representative cross section image from a sciatic nerve from mice fed a CD or HFHFD for 12 weeks staining for Cd31 (vasculature), F4/80 (macrophages), and DAPI (nuclei). b. Quantification of RecMacs (Cd64+Ly6c+) macrophages within the live Cd45+Cd11b+Ly6g- myeloid cell population following a perfusion with PBS in mice fed a CD or HFHFD for 12 weeks. (n = 4 mice in CD and 5 mice in HFHFD group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). c. Representative image from sciatic nerves of Ccl2-RFP mice fed a CD or HFHFD for 12 weeks showing no overlap with PGP9.5 (neurons) (representative from 2 independent experiments, n = 3 mice per group each, scale bar = 100 um). d. Representative image from sciatic nerves of Ccl2-RFP mice fed a CD or HFHFD for 12 weeks showing no overlap with S100 (Schwann Cells) (representative from 2 independent experiments, n = 3 mice per group each, scale bar = 100 um). Source Data

Extended Data Fig. 7. HFHFD feeding does not lead to demyelination or evidence of myelin disturbances after 12 weeks in the sciatic nerve.

a. Representative EM images from sciatic nerves of mice fed a CD or HFHFD for 12 weeks. b. Scatterplot and slope of each quantified axon diameter vs. g-ratio from CD samples and HFHFD samples. (n = 3 mice on CD and 4 mice on HFHFD, data are presented as mean values +/- SEM). c. Quantification of average axon diameter per mouse from 3 non-consecutive images. (n = 3 mice on CD and 4 mice on HFHFD, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). d. Quantification of g-ratio split by axon diameter ranges per mouse from 3 non-consecutive images. (n = 3 mice on CD and 4 mice on HFHFD, data are presented as mean values +/- SEM, p-values from multiple unpaired t-tests corrected using Two-stage step-up method (Benjamini, Krieger, Yekutieli)). e. Quantification of percent myelinated fibers per mouse from 3 non-consecutive images. (n = 3 mice on CD and 4 mice on HFHFD, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). f. Quantification of average number of axons per Remak bundle per mouse from 3 non-consecutive images. (n = 3 mice on CD and 4 mice on HFHFD, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). Source Data

Extended Data Fig. 8. Analysis of skin and dorsal root ganglia immune cells from mice fed a HFHFD or CD for 12 weeks shows no change in immune cell composition.

a. Workflow for tissue collection, processing, sorting, and sequencing of skin immune cells from male mice fed a HFHFD or CD for 12 weeks with 4 mice per group pooled into 2 samples. b. Number of cells analyzed in each sample post-filtering. c. Dot plot of marker genes used to identify different immune cells in dataset. d. UMAP plot showing clusters identified in sciatic nerves of CD and HFHFD fed mice from 2 samples per group. e. Barplot showing proportions of the different cell types per group. f. Flow cytometry analysis of recruited macrophages (Ly6c+Cd64+) within the myeloid cell population (Live Cd45+Cd11b+Ly6g-). (n = 3 mice in CD ad 5 mice in HFHFD group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). g. Volcano plot of differentially expressed genes in skin resident macrophages in HFHFD samples compared to CD, dashed lines at p-value = 0.05 and Log2- Fold Change −0.5 and 0.5. h. GO molecular function analysis of upregulated genes in resident macrophages. i. Representative flow cytometry contour plots from DRG immune cells from CD and HFHFD-fed mice for 12 weeks pre-gated on live Cd45+ Cd11b+ Ly6g- cells showing Ly6c and Cd64 staining. j. Flow cytometry quantification of recruited macrophages (Ly6c+Cd64+) within the myeloid cell population (Live Cd45+Cd11b+Ly6g-). (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). The graphics in a were created with BioRender.com. Source Data

Extended Data Fig. 9. Mice lacking CCR2-KO show no overt difference in glycemic status, T cells, or mechanical thresholds.

a. Body weights from male WT and Ccr2-KO mice fed a HFHFD over time (n = 5 mice per group, data are presented as mean values +/- SEM). b. Glucose Tolerance Test from male WT and Ccr2-KO mice fed a HFHFD for 12 weeks (n = 5 mice per group, data are presented as mean values +/- SEM) c. Area under the curve from glucose tolerance test from male WT and Ccr2-KO mice fed a HFHFD for 12 weeks (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). d. Percent glycated HbA1c from whole blood from male WT and Ccr2-KO mice fed a HFHFD for 12 weeks (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). e. Fasting insulin levels measured from plasma of male WT and Ccr2-KO mice fed a HFHFD for 12 weeks (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). f. Flow cytometry analysis of T Cells (Cd3+) within live Cd45+ immune cells (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). g. Mechanical sensitivity assessed using a von Frey assay in male WT and Ccr2-KO mice fed a HFHFD (n = 5 mice per group, data are presented as mean values +/- SEM, p-values from 2way ANOVA with Sidak multiple comparisons). h. Heat sensitivity assessed using Hargreaves assay in male WT and Ccr2-KO mice fed HFHFD for 24 weeks (n = 5 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test). Source Data

Extended Data Fig. 10. CCR2 blockade has no effect on CD fed mice, leads to increased warm preference after HFHFD and the effect of Lgals3 knockdown is not mediated by reduced recruitment of macrophages to the nerve.

a. Body weights from male WT mice fed a HFHFD over time injected with CVC or Vehicle (n = 4 mice per group, data are presented as mean values +/- SEM). b. Mechanical sensitivity was assessed using the von Frey up down method in mice at 8 weeks on a HFHFD and at 12 weeks following CVC or Veh injections (n = 9 CD and 8 HFHFD, data are presented as mean values +/- SEM). c. Heat sensitivity assessed using Hargreaves assay in mice fed a CD for 12 weeks and injected with CVC or Veh starting after 8 weeks on diet. (n = 7 mice per group, data are presented as mean values +/- SEM). d. quantification of IENFD from mice fed a CD for 12 weeks and injected with CVC or Veh starting after 8 weeks on diet. (n = 7 mice per group, data are presented as mean values +/- SEM). e. Time spent per zone on a thermal gradient ring (n = 5 WT and n = 4 KO female mice, data are presented as mean values +/- SEM) (p-value at 28 °C < 0.0001, p-value at 10 °C = 0.1963, 14 °C = 0.2177, 17 °C = 0.1284, 21 °C = 0.4804, 24 °C = 0.8515, 28 °C < 0.0001, 31 °C = 0.1923, 35 °C = 0.9993, all other p-values >0.9999 from 2way ANOVA with Sidak multiple comparisons). f. Time spent in different temperature zones on thermal gradient ring (n = 4 male mice per group, data are presented as mean values +/- SEM) (p-value at 28 °C = 0.5052, 31 °C = 0.0105, 35 °C = 0.4037, at 39 °C = 0.0135, all other p-values >0.9999 from 2way ANOVA with Sidak multiple comparisons). g. Body weight at start and end of experiment after HFHFD feeding (n = 5 WT and n = 4 KO, data are presented as mean values +/- SEM). h. Quantification of recruited macrophages (recMacs) (Ly6c+Cd64+) from nerves within the live Cd45+Cd11b+ Ly6g- population using flow cytometry or WT or Lgals3-KO mice fed HFHFD for 12 weeks (n = 7 mice per group, data are presented as mean values +/- SEM, p-value from two-tailed unpaired student’s t-test).i. Illustrative summary of the findings described in this manuscript created with BioRender.com. Source Data